The current trend in genome-wide association studies is to identify regions where the true disease-causing genes may lie by evaluating thousands of single-nucleotide polymorphisms (SNPs) across the whole genome. However, many challenges exist in detecting disease-causing genes among the thousands of SNPs. Examples include multicollinearity and multiple testing issues, especially when a large number of correlated SNPs are simultaneously tested. Multicollinearity can often occur when predictor variables in a multiple regression model are highly correlated, and can cause imprecise estimation of association. In this study, we propose a simple stepwise procedure that identifies disease-causing SNPs simultaneously by employing elastic-net regularization, a variable selection method that allows one to address multicollinearity. At Step 1, the single-marker association analysis was conducted to screen SNPs. At Step 2, the multiple-marker association was scanned based on the elastic-net regularization. The proposed approach was applied to the rheumatoid arthritis (RA) case-control data set of Genetic Analysis Workshop 16. While the selected SNPs at the screening step are located mostly on chromosome 6, the elastic-net approach identified putative RA-related SNPs on other chromosomes in an increased proportion. For some of those putative RA-related SNPs, we identified the interactions with sex, a well known factor affecting RA susceptibility.

Background: The development and application of clinical prediction models using machine learning in clinical decision support systems is attracting increasing attention. Objective: The aims of this study were to develop a prediction model for cardiac arrest in the emergency department (ED) using machine learning and sequential characteristics and to validate its clinical usefulness. Methods: This retrospective study was conducted with ED patients at a tertiary academic hospital who suffered cardiac arrest. To resolve the class imbalance problem, sampling was performed using propensity score matching. The data set was chronologically allocated to a development cohort (years 2013 to 2016) and a validation cohort (year 2017). We trained three machine learning algorithms with repeated 10-fold cross-validation. Results: The main performance parameters were the area under the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (AUPRC). The random forest algorithm (AUROC 0.97; AUPRC 0.86) outperformed the recurrent neural network (AUROC 0.95; AUPRC 0.82) and the logistic regression algorithm (AUROC 0.92; AUPRC=0.72). The performance of the model was maintained over time, with the AUROC remaining at least 80% across the monitored time points during the 24 hours before event occurrence. Conclusions: We developed a prediction model of cardiac arrest in the ED using machine learning and sequential characteristics. The model was validated for clinical usefulness by chronological visualization focused on clinical usability.

Purpose: Amnestic mild cognitive impairment (aMCI) is a transitional state between normal aging and Alzheimer’s disease (AD). However, not all aMCI patients are observed to convert to AD dementia. Therefore, developing a predictive algorithm for the conversion of aMCI to AD dementia is important. Parametric methods, such as logistic regression, have been developed; however, it is difficult to reflect complex patterns, such as non-linear relationships and interactions between variables. Therefore, this study aimed to improve the predictive power of aMCI patients’ conversion to dementia by using an interpretable machine learning (IML) algorithm and to identify the factors that increase the risk of individual conversion to dementia in each patient.Methods: We prospectively recruited 705 patients with aMCI who had been followed-up for at least 3 years after undergoing baseline neuropsychological tests at the Samsung Medical Center between 2007 and 2019. We used neuropsychological tests and apolipoprotein E (APOE) genotype data to develop a predictive algorithm. The model-building and validation datasets were composed of data of 565 and 140 patients, respectively. For global interpretation, four algorithms (logistic regression, random forest, support vector machine, and extreme gradient boosting) were compared. For local interpretation, individual conditional expectations (ICE) and SHapley Additive exPlanations (SHAP) were used to analyze individual patients.Results: Among the four algorithms, the extreme gradient boost model showed the best performance, with an area under the receiver operating characteristic curve of 0.852 and an accuracy of 0.807. Variables, such as age, education, the scores of visuospatial and memory domains, the sum of boxes of the Clinical Dementia Rating scale, Mini-Mental State Examination, and APOE genotype were important features for creating the algorithm. Through ICE and SHAP analyses, it was also possible to interpret which variables acted as strong factors for each patient.Conclusion: We were able to propose a predictive algorithm for each aMCI individual’s conversion to dementia using the IML technique. This algorithm is expected to be useful in clinical practice and the research field, as it can suggest conversion with high accuracy and identify the degree of influence of risk factors for each patient.

Importance: Over the past 10 years, treatment of non–small cell lung cancer (NSCLC) has been continually revolutionized. However, standard clinical trials may not reflect current multiple lines of treatment and corresponding outcomes in a timely manner.Objective: To investigate outcomes associated with new treatment of NSCLC in a clinical setting.Design, Setting, and Participants: This cohort study included patients with NSCLC between January 1, 2010, and November 30, 2020, who received any anticancer treatment at Samsung Medical Center in Korea. Data were analyzed from November 2021 through February 2022.Exposures: Clinical and pathological stage, histology, and major druggable sequence variation, including epidermal growth factor receptor (EGFR), anaplastic lymphoma kinase (ALK), ROS1, RET, MET exon 14 skipping, BRAF V600E, KRAS G12C, and NTRK between 2 periods (period I: 2010-2015 vs period II: 2016–2020).Main Outcomes and Measures: The primary outcome was the 3-year survival rate of NSCLC. Secondary outcomes included median overall survival, progression-free survival, and recurrence-free survival.Results: Among 21 978 patients with NSCLC (median [range] age at diagnosis, 64.1 [57.0-71.0] years; 13 624 males [62.0%]), there were 10 110 patients in period I and 11 868 patients in period II; adenocarcinoma (AD) was the predominant histology (7112 patients [70.3%] in period I and 8813 patients [74.3%] in period II). There were 4224 never smokers [41.8%] in period I and 5292 never smokers [44.6%] in period II. Compared with patients in period I, patients during period II were more likely to undergo molecular tests in the AD (5678 patients [79.8%] vs 8631 patients [97.9%]) and non-AD (1612 of 2998 patients [53.8%] and 2719 of 3055 patients [89.0%]) groups. In patients with AD in period I, 3-year survival rates were 92.8% (95% CI, 91.8%-93.7%), 72.4% (95% CI, 68.3%-76.8%), 56.7% (95% CI, 53.4%-60.2%), and 28.7% (95% CI, 27.0%-30.4%) for stage I, II, III, and IV, respectively. In period II, 3-year survival rates of patients with AD were 95.1% (95% CI, 94.4%-95.9%), 82.5% (95% CI, 79.1%-86.1%), 65.1% (95% CI, 61.8%-68.6%), and 42.4% (95% CI, 40.3%-44.7%) for each stage, respectively. In patients without AD, 3-year survival rates were 72.0% (95% CI, 68.8%-75.3%), 60.0% (95% CI, 56.2%-64.1%), 38.9% (95% CI, 35.6%-42.5%), and 9.7% (95% CI, 7.9%-12.1%) for each stage in period I. In period II, the 3-year survival rates of patients without AD were 79.3% (95% CI, 76.3%-82.4%), 67.3% (95% CI, 62.8%-72.1%), 48.2% (95% CI, 44.5%-52.3%), and 18.1% (95% CI, 15.1%-21.6%) for each stage.Conclusions and Relevance: In this cohort study of 10 years of clinical data, survival outcomes were improved across all stages, with larger increases in patients with stage III to IV disease. The incidence of never-smokers and the use of molecular testing increased.